In the category of clear water-based liquids which burn cleanly enough to be used indoors, and which could be confused with water in a testing arrangement (since it would be so unexpected as the 'trick' used to pull-off the deception) - there are several choices.
These are miscible and with 40-50% water and the resultant blend would be combustible at that dilution level - would go undetected by a group of observers who assumed that it was water. All of these ingredients would be expected to be legitimately found in any company which produces or evaluates alternative fuels - and if the ruse was discovered prematurely . "oops, Igor, you brought in the wrong container," or else "yes, our municipal water is very polluted here". A form of alcohol, called Sterno, or 'canned heat' is made from jellied alcohol and burned directly from its can. The food service industry uses it for buffet heating and chafing, and it produces little smell and no monoxide; however, it would need to be used with less gel and no added color - to resemble water. The primary exhaust product is steam (~2/3). The gel keeps alcohol volatility under control and acts to denature it, as it is toxic to ingest. The heat content is about 18 MJ/l - or half of gasoline. Hydrogen peroxide produces only steam. HOOH is more viscous than water, but appears colorless in solution. It is both an oxidant an a propellant. When used in a blend, it would provide free oxygen and steam, so that air is not needed to combust the other ingredients (or less is needed). Ethylene glycol is an automotive antifreeze. In its pure form, it is an odorless, colorless, sweet-tasting liquid and toxic to consume. As antifreeze is denatured to discourage pets from ingesting and color is added. In pure form, it can serve as the gel for the other ingredients. A mix of these with water should be adequate to produce a combustible non-volatile, odorless liquid whose primary exhaust product is steam, with no remnant smell and with a heat content sufficient to convert all of the liquid to steam using a common catalyst. This hypothetical liquid would be more viscous than water, and with a peristaltic pump, it would be expected to produce more of an abnormal sloshing noise than would be expected of water. In summary - the heat content of this water-like liquid would be under 10 MJ/l and the net effect of the deception will be to convert all of it into steam so there is absolutely no argument that a chemical, non-nuclear reaction is not sufficient to produce the steam which is documented. There will be pure CO2 present in the exhaust, but it would go unnoticed, as it is odorless and colorless.